Strojniki vestnik - Journal of Mechanical Engineering 57(2011)2, 100-109

DOI:10.5545/sv-jme.2009.018

Paper received: 27.01.2009

Paper accepted: 10.09.2010

System Approach to Vehicle Suspension System Control

in CAE Environment
Vladimir Popovic1,* - Branko Vasic1 - Milos Petrovic2 - Sasa Mitic1
1 University of Belgrade, Faculty of Mechanical Engineering, Serbia
2 Institute for Research and Design in Commerce & Industry, Serbia
In recent years, motor vehicles industry has shown a tendency of replacing electromechanical
components by mechatronic systems with intelligent and autonomous properties. The integration of
hardware components and implementation of advance control function characterize this replacement. In
this paper, we have applied the system approach and system engineering methods in the initial phase
of vehicle active suspension development. An emphasis has been placed upon the interrelations between
computer-aided simulation and other elements of the development process. The benefits of application of
active suspension simulation are numerous: reduction of time to market, the new and improved functions
of mechatronic components/devices, as well as the increased system reliability. In suspension model
development, we used CAD/CAE tools, as well as the multipurpose simulation programs. For simulation,
we used the one-quarter vehicle model. The modelling was carried out through the state-space equation,
after which we designed the controller for our system. During this, we considered only the digital systems
of automatic regulation.
2011 Journal of Mechanical Engineering. All rights reserved.
Keywords: active suspension system, system approach, simulation, control system, PID controller

0 INTRODUCTION
Considerable effort aiming at reducing
the cost, the ever increasing expectations of
users when it comes to reliability, the greater
complexity of modern technical systems, and
other requests, bring us to the conclusion that the
approach based on system engineering principles,
remains the only way to stay competitive on the
market. Intensifying the integration of electronic
components into areas which used to be strictly
mechanical and the degree of change within
technologies, all result in increased demands for
dealing with the problems on the basis of what is
offered by system engineering. In recent years,
motor vehicles industry has started developing
an unusual trend - the mechatronic systems with
intelligent and autonomous abilities. Development
of integrated mechatronic systems should, in
its own way, play the key role in motor vehicles
industry [1]. Mechatronics is being more and
more accepted as the design methodology for
all motor vehicle systems. Verification of proper
functionality through simulation can improve
reliability and reduce component design time.
Replacement of simple electro-mechanical
100

components by intelligent mechatronic systems

is characterised by two aspects: integration
of hardware components and implementation
of advanced control functions (functional and
algorithm integration). Hence, mechanical
components are operated through basic sensor
feedback of low level and intelligent processing of
advanced level information.
The goal in active suspension control
research is to improve the ride performance,
generally quantified by sprung mass acceleration,
while maintaining an acceptable level of
suspension stroke and tyre deflection as packaging
and handling measures. Ride comfort, road
holding ability and suspension deflection are the
three main performance criteria in any vehicle
suspension design [2]. Finally, all control concepts
aim at introducing additional forces to the
suspension system to reduce roll, pitch, and stroke
movements, as well as body and wheel vibrations.
Therefore, the basic control approach is similar
for all actuation systems. Then, the physical
structure of the considered system determines the
further controller design and also directly limits
the achievable systems performance [3].

*Corr. Authors Address: University of Belgrade, Faculty of Mechanical Engineering, Kraljice Marije 16,
11120 Belgrade, Serbia, vpopovic@mas.bg.ac.rs

Strojniki vestnik - Journal of Mechanical Engineering 57(2011)2, 100-109

1 DESIGN DEVELOPMENT PROCESS

The design procedure, which is to be
carried out in the case of mechatronic systems,
is very demanding - it is necessary that system
engineering, within the field of mechanics,
electronics and computer science, forms a
completely integrated system. Hence, intelligent
testing technologies, supported by CAE
(Computer Aided Engineering), which provide
numerical simulation models, are to be employed
during component development and their
qualification. An emphasis should be placed upon
the interaction between computer aided system
simulation and experimental testing techniques
through intelligent information processing. CAE
is a technology that enables computer analysis
of the design, created within CAD (Computer
Aided Design) technology. By the application
of CAE technology it is now possible to have a
much better linking of design, testing and design
improvement, which are the development phases
that have, so far, been almost separate. Fig. 1
shows a diagram of design development process
[4]. After theoretical modelling and experimental
system identification, the future design steps
are control system analysis and model based
controller design. The demands for efficient
optimisation and testing include:
Software development environment with
accurate, firm connections among different
CAE tools, such as CAD, FEM and MBS
programs, and CACE (Computer Aided
Control Engineering) tools. An example
of such an environment is shown in Fig. 2
[5]. On the basis of this diagram, we can
determine the role of Matlab (and Simulink)
in the framework of the complete process of
modelling and simulation of the new technical
system.
Experimental, high performance platform
which, in combination with software
development
environment,
completely
automates the implementation process from
off-line simulation to engaging complex
controllers with real time operation.
Interactive
testing
of
mechatronic
components, using advanced hardware simulators
in feedback, plays an important role. The idea is
to build mechatronic components into a virtual

environment in which vehicle movement, external

load and the surrounding mechanical systems,
are simulated in real time. A complex model of
vertical vehicle dynamics has been developed
using CAE tools and multi-purpose simulation
programs. On the basis of reduced models, a
multivariable, powerful control system should be
designed for active suspension control. Fig. 3 [4],
[5] shows the corresponding vehicle model with
a passive spring that bears static load, which is to
be used for suspension system simulation.

Fig. 1. Design development process

Fig. 2. Software development environment with

connections among different CAE tools
2 ACTIVE SUSPENSION SYSTEM
MODELLING
Various approaches have been proposed
to improve the performance of active suspension
designs, such as linear optimal control, fuzzy logic
and neural network control, adaptive control, H
control, nonlinear control, LQG control, skyhook
control algorithm, etc [2] to [8]. Also, many
approaches are presented to deal with the multiobjective requirement of vehicle suspensions.
In particular, H control of active suspensions is
intensively discussed in the context of robustness
and disturbance attenuation [2]. An interesting
solution is a fuzzy sliding mode controller to

System Approach to Vehicle Suspension System Control in CAE Environment

101

Strojniki vestnik - Journal of Mechanical Engineering 57(2011)2, 100-109

control an active suspension system and evaluate

its control performance [9] to [12]. The fuzzy
sliding mode controller employed the error of
the sprung mass position and the error change to
establish a sliding surface, and then introduced
the sliding surface and the change of the sliding
surface as input variables of a traditional fuzzy
controller in controlling the suspension system
[13]. One of the possible solutions is vibration
control of a vehicle active suspension system
using a new robust neural network control system
[14]. It consists of a robust feedback controller and
feedforward neural network predictive controller.

Fig. 3. 1/4 vehicle model

In the design of the suspension system,
the following is used: vehicle model [2] to
[8], [11] and [13], vehicle model [15] and full
vehicle model [14] and [16]. We used vehicle
model (Fig. 3) to simplify the problem to a onedimensional spring-damper system. The vehicle
represents a complex oscillatory system with
a great number of freedom degrees. Vehicle
oscillations are caused mainly by road disturbance.
In reviewing vehicle oscillations we adopted the
following starting points:
the vehicle is in the rectilinear motion at a
constant speed;
the wheels are always in contact with the road
and that is a one-spot contact;
road disturbance is the same on the left and
the right wheel and the car is symmetrical
relative to the longitudinal axis;
mass distribution coefficient is approximately
1.
Although the real characteristics of the
suspension system (k1, b1) are non-linear, we
adopted constant values for them in this paper
which, to a certain extent, enable the linearization
102

of the model. Also, the step function that we

used for simulation has certain limitations since
if the tyre of the vehicle is excited by the step
function, the tyre will jump and the contact with
the driving surface will be lost. Furthermore,
when compressed, damping of the suspension
damper is many times smaller than damping when
the damper is extended. These limitations will
be further discussed in the conclusion. The data
that we used during the simulation refer to the
bus. The simulation was performed using Matlab
interactive environment. The designations in Fig.
3 have the following meanings [4] and [5]:
body mass (m1) = 2250 kg;
suspension mass (m2) = 290 kg;
spring constant of suspension system
(k1) = 72000 N/m;
spring constant of wheel and tyre
(k2) = 450000 N/m;
damping constant of suspension system
(b1) = 315 Ns/m;
damping constant of wheel and tyre
(b2) = 13500 Ns/m;
control force (Fa) = force from the controller
we are going to design.
When the vehicle is experiencing any
road disturbance, the body should not have large
oscillations, and the oscillations should dissipate
quickly. This, at the same time, is our principal
task. Since the distance x1-Z is very difficult to
measure, and the deformation of the tyre x2-Z is
negligible, we will use the distance x1-x2 instead
of x1-Z as the output in our problem. The road
disturbance Z will be simulated by a step input.
This step could represent the vehicle coming
out of a pothole. We wish to design a feedback
controller so that the output x1-x2 has an overshoot
less than 5% and a settling time shorter than 5 sec.
Dynamic properties of the system can, in time
range, be most suitably defined by the values of
parametres that determine the system response
(overshoot and settling time) [17].
2.1 Model of the System in State-Space
The assessment of the quality of automatic
regulation of a system behaviour essentially
boils down to estimating the error between a
predetermined value and the real value of the
controlled variable. The knowing of this error at

Popovic, V.- Vasic, B. - Petrovic, M. - Mitic, S.

Strojniki vestnik - Journal of Mechanical Engineering 57(2011)2, 100-109

any point would give complete information on

the features of the observed system. Due to the
diversity of the laws of the system input change,
which might occur in the normal work regime,
such an approach, based on estimating the current
values of error, is inappropriate speaking from
the aspect of practice. Therefore, it is preferable
to make an estimation of the relevant system
characteristics on the basis of the features it
manifests while being disturbed by the typical
input signals [17]. On the basis of Fig. 3 and
Newtons Law, the following dynamic equations,
which represent the mathematical model of the
dynamic system [4] and [5] are obtained:
m1
x1 = -b1 ( x1 - x2 ) - k1 ( x1 - x2 ) + Fa
m2
x2 = b1 ( x1 - x2 ) + k1 ( x1 - x2 ) +
+b ( Z - x ) + k ( Z - x ) - F .
2

x2 =
+

b1 k1
b
Y1 +
Y1 + 2 ( Z - x2 ) +
m2
m2
m2
F
k2
( Z - x2 ) - a .
m2
m2

b1 b1 k1 k1
+ )Y1 - ( + )Y1 m1 m2
m1 m2

(5)

k
b2
1
1
( Z - x2 ) - 2 ( Z - x2 ) + Fa ( + ).
m2
m1 m2
m2

Since we cannot use second derivatives in

the state-space representation, we integrate this
Eq. to get Y1 :
b
b
b
Y1 = -( 1 + 1 )Y1 - 2 ( Z - x2 ) +
m1 m2
m2
+ (-(

k1 k1
k
+ )Y1 - 2 ( Z - x2 ) +
m1 m2
m2

+ Fa (

(2)

F
b1 k1
Y1 - Y1 + a ,
m1
m1
m1

x1 -
x2 = Y1 = -(

(1)

When addressing new ways for system

analysis and synthesis that avoid the problem of
solving algebraic and differential equations, the
system analysis and synthesis in state-space
mainly come to mind. Representation of the
system in the form of state-space equation is much
easily derived from differential equations than by
using Laplace transformations. To be a valid statespace representation, the derivative of all states
must be in terms of inputs and the states
themselves. Now, let us choose the states that we
shall be using. Firstly, let us divide the Eqs. (1)
and (2) by m1 and m2, respectively and introduce
the substitute Y1 = x1-x2. Note that Z appears in
the Eq. for x 2 :

x1 = -

should be. Therefore, we subtract Eq. (4) from Eq.

(3) to get an expression for Y1 :

(6)

1
1
+ ))dt.
m1 m2

No derivatives of the input appear in this

equation, and Y1 is expressed in terms of states
and inputs only, except for the integral. Let us call
the integral Y2. Assuming that x2 = x1 - Y1, from
the Eq. (6) we get the state-space equation for Y1:
b
b
b
Y1 = -( 1 + 1 )Y1 - 2 ( Z - x1 + Y1 ) + Y2 .
m1 m2
m2

(7)

Then we shall substitute the derivative of

Y1 into the Eq. (3) of the derivative of x1:

x1 = +

b1b2
b b
b
b
k
x1 + ( 1 ( 1 + 1 + 2 ) - 1 )Y1 +
m1m2
m1 m1 m2 m2 m1

(8)

F b
b1b2
Z + a - 1 Y2 .
m1m2
m1 m1

(3)

(4)

The first state-space equation will be x1.

Since no derivatives of the input appear in the
equation for x1 , we choose x1 for the second
state. Then, we choose the third state as the
difference between x1 and x2. After doing some
algebra, we will determine what the fourth state

Fig. 4. Open-loop response for the unit step

actuated force

System Approach to Vehicle Suspension System Control in CAE Environment

103

Strojniki vestnik - Journal of Mechanical Engineering 57(2011)2, 100-109

The state variables are x1, x1 , Y1 and Y2.

The matrix from the above Eq. is:
0
bb
x1 1 2

m1m2
x1 = b
Y1 2
m2
Y2 k
2

m2
0
1

m1
+
0

1 + 1
m1 m2

1
0

0
k
b1 b1 b1 b2
( +
+ ) 1
m1 m1 m2 m2 m1

b1 b1 b2
+
+ )
m1 m2 m2

k1 k1 k2
+
+ )
m1 m2 m2

0
b1
x1
m1
* x1 +
1 Y1

Y2
0

0
b1b2
m1m2
Fa ,
b
2 Z
m2

k
2
m2

x1
x
F
Y = [ 0 0 1 0] 1 + [ 0 0] a .
Y1
Z

Y2

(9)

(10)

We can put the above state-space Eqs. (9)

and (10) into Matlab by defining the four matrices
of
the
standard
state-space
equation:
=
X AX + BZ ; Y = CX + DZ. Thus, a new m-file,
formed on the basis of state-space equation has
been created. By adding step(A,B,C,D,1)
command into the m-file and its activating, in
Matlab command window an open-loop response
for the unit step actuated force is obtained.
Generally, new functions to the Matlab vocabulary
are added by expressing them in terms of existing
functions. The existing commands and functions
that compose the new function reside in a text file
called an m-file [18]. M-files can be either scripts
or functions. Scripts are simply files containing a
sequence of Matlab statements. Functions make
use of their own local variables and accept input
arguments. The name of an m-file begins with an
alphabetic character and has a filename extension
of .m.
Fig. 4 shows that the system is
underdumped. People sitting on the bus will feel
a small amount of oscillation and the steadystate error is about 0.01 mm. However, the
bus needs unacceptably long time to reach the
steady state the settling time is rather long.
The solution to this problem lies in including a
feedback controller into the block diagram of
the system. By the term controller we also imply
104

the controller and the actuator. By adding the

command step(A,0.1*B,C,D,2) into the m-file
the open-loop response to 0.1 m step disturbance
is obtained.

Fig. 5. Open-loop response to 0.1 m step

disturbance
Fig. 5 shows that, when the bus passes a
10 cm high bump on the road, the bus body will
oscillate for an unacceptably long time ( 50
sec), and with a much larger amplitude than the
initial impact. The big overshoot and the slow
settling time will cause damage to the suspension
system. As already stated above, the solution to
this problem is to add a feedback controller into
the system to improve performance. The block
diagram of the thus obtained closed-loop system
is presented in Fig. 6.

Fig. 6. Closed-loop system block diagram

3 SYNTHESIS OF ACTIVE SUSPENSION
DIGITAL SYSTEM
Apart from the PID algorithm, which was
used for controlling an active vehicle suspension
system, many modern solutions for control
systems and actuators themselves are found in the
literature:
H control with actuator time delay [2];

Popovic, V.- Vasic, B. - Petrovic, M. - Mitic, S.

Strojniki vestnik - Journal of Mechanical Engineering 57(2011)2, 100-109

Impedance control for the vehicle suspension

system by electro hydraulic actuator the
electro hydraulic actuator comprises a servovalve and a hydraulic cylinder [6];
Hybrid control technique applied to a vehicle
active suspension system using skyhook and
adaptive neuro active force control [8]. The
overall control system essentially comprises
four feedback control loops; the innermost
proportional-integral control loop for the
force tracking of the pneumatic actuator, the
intermediate skyhook and active force control
control loops for the compensation of the
disturbances and the outermost PID control
loop for the computation of the optimum
target/ commanded force. Pneumatic actuators
have the advantage of low cost, a high power
to weight ratio, ease of maintenance and a
readily available and cheap power source;
Different kinds of rotational actuators,
electric levelling actuators, electromechanical
actuators and electric damper actuators [7];
The optimization technique used for the
optimization of the controller parameters
and the spring rates is based on a genetic
algorithm [7]. The main advantage of this
method and all other non-gradient based
optimization methods is that it does not need
the derivatives of the objective function,
which are difficult to calculate from a numeric
simulation model;
Knowledge-based fuzzy logic controllers,
variable structure controllers, backstepping
controllers [16];
Possible solutions are dampers taking
advantage of the electro-rheological or
magneto-rheological property of liquids. Both
principles are based on the alteration of the
damping mediums viscosity depending on
an applied electric or magnetic field. These
systems are mainly applied with semi-active
vehicle suspensions [3] and [19];
Somewhat older solutions are hydraulic,
hydro-pneumatic, electro-mechanical and
pneumatic systems. Most of these systems
also comprise passive elements such as
conventional springs and dampers or
hydro-pneumatic damping and springing
components involved in the actuation system
[3];

In the integrated intelligence type

implementation, actuator intelligence and
actuation are combined into a single material.
A piezoceramic actuator is a typical example
of this type [20].
Actuators may have undesirable static
and dynamic characteristics introducing many
problems for the PID control loop. For instance,
limit cycles caused by actuator backlash cannot be
removed by adjusting PID parameters. Therefore,
a nonlinearity compensation algorithm using an
inverse model and a simple position controller
are implemented in order to maintain the actuator
characteristics to be as linear as possible within
the attainable actuation range [20]. The PID
controller has been most widely employed because
of its simple structure and the effective use in
industry [21] and [22]. Despite many advances
for the PID controller, this structure has constant
gain parameters and is not good for decreasing
velocity control error [14]. For the purpose of our
paper, the simplest conventional PID controllers
are obtained through an easy discretization of
analogous equivalents. By the application of
z-transformation, the function of incremental PID
controller discrete transfer in the following form
is obtained:
Dp ( z ) =

U ( z)
T / TI TD

= K 1 +
+ (1 z 1 )
1
E( z)
T
1 z

(11)

or
Dp ( z ) = K P +

KI
+ K D (1 z 1 ) ,
1 z 1

where KP = K, Ki = KT / Ti and KD = KTD / T are

known as P-, I- D-action factors, respectively (K
is the factor of gain; TI and TD are time constants
of integral and differential action).
The structure of the system with a digital
PID controller is given in Fig. 7 [17]. Note that
without the block with broken lines in Fig. 7a, the
regulation contour contains an incremental PID
controller. With this block, however, this contour
becomes equivalent to the contour in Fig. 7b,
which contains a positional PID regulation law.
The positional type of digital law seems natural;
however, the incremental kind is more readily
used. Incremental algorithm must be used if it is
immediately connected to the executive organ of
the stepping motor type, and it is also convenient

System Approach to Vehicle Suspension System Control in CAE Environment

105

Strojniki vestnik - Journal of Mechanical Engineering 57(2011)2, 100-109

for it works easily even in the case when the

executive organ output is in border line position.
Also, it is important to emphasize that there are
numerous methods of on-hardware optimization
and improvement of the performance of openloop stepping motor system dynamics [23]. The
principal defect of incremental algorithm is not
being able to see the position of executive organ
output. Hence, when proportional executive or
integrating (but included by feedback) organ is
used, the positional form of PID algorithm must be
used, or the incremental form must be converted
into the positional, which essentially boils down
to implementation of the block with broken
lines in Fig. 7a. By adopting a smart actuator
technology, standard PID control techniques can
be used successfully in the presence of undesirable
actuator characteristics. The smart actuator could
be applied to an on-line PID controller retuning
using a standard pole-placement technique to
counteract degraded actuator performance [20].

sampling time T. This is a highly important step,

since road disturbance very quickly influences the
output. Because the controller can see only the
consequence of the disturbance after a complete
sampling time, we shall have to choose a sampling
time T, short enough, so that the (x1 - x2) output
will not exceed the initial requirement of 5% in
one selection time. To choose a selection time,
we should carefully consider the initial part of
the system response graph. Fig. 5 tells us that
the output soon assumes negative values, and
then starts to oscillate. We shall simulate only the
beginning of this response by adjusting the time
vector in 0 to 0.005 range. The response to 0.1 m
step disturbance is simulated by multiplying B
matrix by 0.1. Into the already formed m-file we
added the command step(A,.1*B,C,D,2,0:0.0001:
.005) and got the open loop response to 0.1 m step
disturbance. This graph shows that the spring with
the deflection rate k1 compresses rather quickly,
and that our system exceeds the initial requirement
of 5 mm response to 0.1 m step disturbance after
only a little more that 0.001 sec. Therefore, we
adjusted T to 0.0005 sec to give the controller
a chance to react. After we have chosen the
sampling time, we converted the system into the
discrete form. We can use Matlab to convert the
above presented state-space model of the system,
using A, B, C and D matrices, into a discrete
state-space model (using Ad, Bd, Cd and Dd) by
c2dm command. This command normally has
six arguments: four space matrices, sampling time
T and the type of circuit hold. In this example we
used the zero-order hold -zoh. Let us add the
following commands into our m-file:
T = 0.0005;
[Ad Bd Cd Dd] = c2dm(A,B,C,D,T,'zoh').

Fig. 7. Structure of the system with a digital PID

controller; a) of incremental type, b) of positional
type
Our task is to design a digital controller, in
state-space, for suspension system control. First
we convert a continual into discrete model, and
then use pole-placement method, as one of the
possible solutions, in order to design a controller.
The state-space model is given in Eqs. (9) and
(10). The first step in converting the continual
into its discrete equivalent is the choice of suitable
106

In this way, we get the response that

represents a new discrete state-space model of
the system. Also, in active control of vehicle
suspension systems, the time delay of the
system is another important issue that needs
careful treatment to avoid poor performance
or even possible instability of the closed-loop
system. Unavoidable time delays may appear
in the controlled channel, particularly in the
digital controller as it carries out the calculations
associated with complex sophisticated control
law, and in sensors and actuators hardware such as

Popovic, V.- Vasic, B. - Petrovic, M. - Mitic, S.

Strojniki vestnik - Journal of Mechanical Engineering 57(2011)2, 100-109

hydraulic actuators where the delays are taken by

the actuators to build up the required control force
[2]. Although the delay time may be short, it can
nevertheless limit the control performance or even
cause the instability of the system when the delay
appears in the feedback loop.

0
0
1.0000
0
1.0000
0
.
.
0
0
1
0000
0
0005

Ada = 0
0.0034 1.0000 0.0125 0.00001 ,

0.0232
0
0.9762 0.0005
0
0.7652 0.0002 0.9051 0.9998
0
0
0
0
0

Bda = 0 0.0034 ,

0 0.0232
0 0.7652

Cda = 1.0e-003 * (0.5000 0 0 0.2500 0) ,

Dda = 0

Fig. 8. Closed-loop response to 0.1 m step

disturbance
3.1 Adding an Integrator
The next step is adding an integrator
into the system, so that the system response in
stationary state is zero. We will add this integrator
in series with the plant. This will result in adding of
the other system states. We shall add an integrator
by representing it in state-space and by using the
series command. This command takes A, B, C
and D matrices of the two systems, links them in a
serial connection as arguments and brings us back
a new set of A, B, C and D matrices. The integrator
in discrete state-space may be represented in any
time T in the following way:
x(k + 1) = x(k ) + Tu (k )
y (k ) = x(k ) +

and

T
u (k ) .
2

(12)

To do this, let us add the following

command into the m-file:
Ai = 1;

Bi = 1; Ci = T; Di = T/2;
[Ada Bda Cda Dda] =
= series(Ad,Bd,Cd,Dd,Ai,Bi,Ci,Di).

Matlab will respond to us with a new set of

matrices, which are the result of integrator action,
with matrix dimension 5 instead of the earlier
dimension 4.

0.

Unfortunately, the output of this equation

is a new integral value. The matrix output Cda
must be changed, in which case we get:
Cda=[Cd 0] Cda = 0 0 1 0 0 .
The controller structure is similar to that of
the state-space controller with continual systems.
We shall now use the place command, so as to
calculate the gain matrix K, which will provide the
desired poles of the closed-loop regulation system.
First, we should decide where to position the
closed-loop poles. Since we need to determine the
position of all the five closed-loop poles, we must
be very selective when it comes to this problem.
To be more precise, we can define the poles so
that they cancel all the system zeros, and at the
same time give us the desired response. Firstly,
we should track all the zeros by converting digital
state-space equations into the transfer function,
and then by determining the square root of the
numerator. We shall use the ss2tf command
which takes state-space matrices and the chosen
input as arguments, and gives us the transfer
function of the numerator and denominator as
the output. Let us add the following code into the
m-file:
[num,den]=ss2tf (Ad, Bd, Cd, Dd, 1);
zeros=roots(num)
The following response is obtained:

zeros = 0.9987 + 0.0065i
0.9987 - 0.0065i
-0.9929 .

System Approach to Vehicle Suspension System Control in CAE Environment

107

Strojniki vestnik - Journal of Mechanical Engineering 57(2011)2, 100-109

We shall select these three zeros as

three out of five desired poles of the closedloop system. One of the remaining two will be
selected at 0.9992, for it is the pole location after
approximately 10000 sampling times (or 5 sec).
The last pole will be selected with z = 0.2, for this
is quick enough to be neglected. Let us add the
following code into our m-file:
p1=.97+.13i; p2=.97-.13i;
p3=-.87;
p1=zeros(1); p2=zeros(2);
p3=zeros(3);
p4=.9992; p5=.5;
K=place(Ada, Bda*[1;0], [p1 p2 p3 p4 p5])
Matlab gives us the following response:
place: ndigits = 15
K = 1.0e+008 *
* (0.0082 0.3458 0.0054 9.8881 0.0096) .
We shall use dstep command to simulate
the closed-loop response. Since multiplying the
state-space vector by K in our controller, gives
one signal only, it is necessary to add a series of
zeros to K vector by multiplying it by [1 0]T. We
shall perform the simulation by the negative value
of the step disturbance -0.1 m, in order to get
the positive value of the deflection for practical
reasons. Let us add the following code:
yout=dstep(Ada-Bda*[1 0]'*K,
-.1*Bda, Cda, -.1*Dda, 2, 10001);
t=0:.0005:5;
stairs(t,yout) .
Fig. 8 shows the closed-loop response to
0.1 m step disturbance. The overshoot is less than
5 mm, and the settling time is not more than 5 sec.
4 CONCLUSION
Over the past few years, active suspension
system has seen increasing application. The
simulation of this system in the initial development
phase is of a multiple importance: i) time is gained
by not having to wait for exploitation results;
ii) it is cost-effective an expensive model in
the lab does not need to be designed and complex
laboratory examinations need not be performed;
iii) suspension systems of almost all categories
and types of motor vehicles can be simulated,
which would, be hard to achieve in laboratory
conditions. Before the practical implementation of
108

the active suspension system, it is recommended

to simulate the obtained analytical solutions on
the computer in order to examine the range of
change of regulative action and the response of
the regulated variable to a typical input signal,
and thus confirm the possibility of the system
physical realization, bearing in mind the actuator
limitations, as well as check whether other possible
limitations within the object of control as a whole
have been reached. The above statements by no
means suggest that laboratory testing is no longer
necessary, but that it can be carried out in a shorter
period of time and with reduced cost [24]. The
key challenge associated with active suspension
and their actuators is the size, weight and energy
consumption required to achieve acceptable
performance. For this reason, the physical
properties of the actuator should be included in the
optimization problem. In practice, actuator faults
are quite often the largest source of control system
degradation. Suspension system modelling was
performed on vehicle model, using the Matlab
interactive environment and by the state-space
equation. The step input is the unit step function,
that is, a certain value of road disturbance. It has
been concluded that after encountering any kind
of obstacle, the settling time and overshoot of
the vehicle, are too long, and that a controller
must be introduced into the suspension system.
We designed a digital controller by the poleplacement method, which makes only one of the
possible solutions. The presented dynamic model
is only a very rough representation of the true
dynamic system, which is applicable only in the
early concept design phases of the research and
development process. It is expected that, in the
near future, this design and the above-mentioned
modifications and the necessary improvements,
would be used in designing such systems in the
motor vehicles industry of our country.
5 ACKNOWLEDGEMENT
This paper is a part of a project of
The Ministry of Science and Technological
Development of Serbia (project number
TR035045 - Scientific-Technological Support
to Enhancing the Safety of Special Road and Rail
Vehicles).

Popovic, V.- Vasic, B. - Petrovic, M. - Mitic, S.

Strojniki vestnik - Journal of Mechanical Engineering 57(2011)2, 100-109

6 REFERENCES
[1] Bolton, W. (2008). Mechatronics - electronic
control systems in mechanical and electrical
engineering, 4th ed. Pearson Prentice Hall.
[2] Du, H., Zhang, N. (2007). H control of
active vehicle suspensions with actuator
time delay. Journal of Sound and Vibration,
vol. 301, p. 236-252.
[3] Fischer, D., Isermann, R. (2004).
Mechatronic semi-active and active vehicle
suspensions. Control Engineering Practice,
vol. 12, p. 1353-1367.
[4] Popovic, V., Jankovic, D., Vasic, B. (2000).
Design and simulation of active suspension
system by using Matlab. Proceedings of
FISITA World Automotive Congress, paper
F2000G269, Seoul.
[5] Popovic, V., Vasic, B., Jankovic, D. (2002).
Development of semi-active suspension
system in CAE environment. Proceedings of
FISITA World Automotive Congress, paper
F02V062, Helsinki.
[6] Fateh, M.M., Alavi, S.S. (2009). Impedance
control of an active suspension system.
Mechatronics, vol. 19, p. 134-140.
[7] Jonasson, M., Roos, F. (2008). Design and
evaluation of an active electromechanical
wheel suspension system. Mechatronics,
vol. 18, p. 218-230.
[8] Priyandoko, G., Mailah, M., Jamaluddin, H.
(2009). Vehicle active suspension system
using skyhook adaptive neuro active force
control. Mechanical Systems and Signal
Processing, vol. 23, p. 855-868.
[9] Huang, S.J., Chen, H.Y. (2006). Adaptive
sliding controller with self-tuning fuzzy
compensation for vehicle suspension
control. Mechatronics, vol. 16, p. 607-622.
[10] Sharkawy, A.B. (2005). Fuzzy and
adaptive fuzzy control for the automobiles'
active suspension system. Vehicle System
Dynamics, vol. 43, no. 11, p. 795-806.
[11] Taskin, Y., Hacioglu, Y., Yagiz, N. (2007).
The use of fuzzy-logic control to improve the
ride comfort of vehicles. Strojniki vestnik Journal of Mechanical Engineering, vol. 53,
no. 4, p. 233-240.
[12] Guclu, R. (2004). The fuzzy-logic control of
active suspensions without suspension-gap

[13]

[14]

[15]

[16]

[17]
[18]
[19]

[20]

[21]

[22]
[23]
[24]

degeneration. Strojniki vestnik - Journal of

Mechanical Engineering, vol. 50, no. 10, p.
462-468.
Lin, J., Lian, R.J., Huang, C.N., Sie, W.T.
(2009). Enhanced fuzzy sliding mode
controller for active suspension systems.
Mechatronics, vol. 19, p. 1178-1190.
Eski, I., Yldrm, S. (2009). Vibration
control of vehicle active suspension system
using a new robust neural network control
system. Simulation Modelling Practice and
Theory, vol. 17, p. 778-793.
Lin, J.S., Huang, C.J. (2004). Nonlinear
backstepping active suspension design
applied to a half-car model. Vehicle System
Dynamics, vol. 42, no. 6, p. 373-393.
Yagiz, N., Hacioglu, Y. (2008). Backstepping
control of a vehicle with active suspensions.
Control Engineering Practice, vol. 16, p.
1457-1467.
Stojic, M.R. (1998). Digital control systems,
4th ed. Faculty of Electrical Engineering
Belgrade. (in Serbian)
Mathworks (2008). http://www.mathworks.
com, accessed on 2008-11-29.
Luo, A.C.J., Rajendran, A. (2007). Periodic
motions and stability in a semi-active
suspension system with MR damping.
Journal of Vibration and Control, vol. 13,
no. 5, p. 687-709.
Lee, D., Allan, J., Thompson, H., Bennett, S.
(2001). PID control for a distributed system
with a smart actuator. Control Engineering
Practice, vol. 9, p. 1235-1244.
Cominos, P., Munro, N. (2002). PID
controllers: recent tuning methods and
design to specification. IEE Proceedings Control Theory and Applications, vol. 149,
no. 1, p. 46-53.
Astrom, K.J., Hagglund, T. (1995). PID
controllers: theory, design and tuning, 2nd
ed. Instrument Society of Automation, USA.
Rogers, J., Craig, K. (2005). On-hardware
optimization of stepper-motor system dyna
mics. Mechatronics, vol. 15, p. 291-316.
Curovic, D., Vasic, B., Popovic, V., Curovic,
N. (2008). Expert planning of production.
Journal of Institute for Research and Design
in Commerce & Industry, vol. 20, p. 49-58.
(in Serbian)

System Approach to Vehicle Suspension System Control in CAE Environment

109